Automatic device and method for detecting meter buzzer sounds

Abstract

To provide an automatic meter buzzer sound detection device that can determine the sound with higher accuracy and ease of use. [Solution] A signal of either the first expected value data or the second expected value data is input to the meter buzzer of the object to be measured, the sound of the meter buzzer is acquired as recording data, the recording data is processed to match the length of the first expected value data to become the first processed recording data, and the recording data is processed to match the length of the second expected value data to become the second processed recording data, the first processed recording data and the second processed recording data are subjected to noise reduction processing, the first correlation coefficient between the first processed recording data and the first expected value data is obtained, the second correlation coefficient between the second processed recording data and the second expected value data is obtained, and a determination is made based on the thresholds of the first and second correlation coefficients to determine whether the first or second processed recording data matches the first expected value data, matches the second expected value data, or is data different from the first and second expected values.

Description

[Technical Field]

[0001] The present invention relates to an automatic determination device and method for determining whether the sound of a meter buzzer used in a vehicle or the like conforms to a design value. [Background technology]

[0002] In the automatic detection technology for meter buzzer sounds used in vehicles, a one-to-one comparison was made between expected value data of the meter buzzer sound based on a predetermined design and recorded data of the meter buzzer sound to be detected. The correlation coefficient between the expected value data and the recorded data was calculated, and the detection was performed based on whether the meter buzzer sound to be detected sounded as expected. Here, the meter buzzer sound to be detected is, for example, the sound emitted when a signal is input to activate the meter buzzer, such as when a seat belt is not fastened, in a prototype meter buzzer.

[0003] Patent Document 1 discloses a broadcast identification device that identifies broadcasting stations and broadcast times based on recorded sound. Patent Document 1 discloses that the audio is determined to match if the correlation coefficient obtained from calculating the correlation coefficient between data A and data B, which are audio signals along the flow of time, is 0.9 or higher. [Prior art documents] [Patent Documents]

[0004] [Patent Document 1] Japanese Patent Publication No. 2008-236548 [Overview of the Initiative] [Problems that the invention aims to solve]

[0005] However, in automatic meter buzzer sound detection technology, there was a problem in that the reliability of the detection was low when comparing data with the same frequency but different periods, specifically when comparing processed expected value data and recorded data on a one-to-one basis.

[0006] Therefore, the present invention aims to provide an automatic determination device and method for determining whether a meter buzzer sound conforms to a design value more easily and with higher accuracy than conventional methods when comparing data with the same frequency but different periods in an automatic determination technology for meter buzzer sounds. [Means for solving the problem]

[0007] An automatic meter buzzer sound detection device that solves the above problem is: An automatic meter buzzer sound determination device that inputs a signal to sound a confirmation buzzer sound to a meter buzzer under measurement, and determines whether the meter buzzer sound sounded by the meter buzzer matches the designed expected value sound data, The automatic determination device comprises a control unit, a storage unit, a microphone, an input unit, a pre-processing unit, and a determination unit. The control unit is The system is configured to create expected value audio data as first expected value data consisting of audio at a predetermined frequency and period, and second expected value data consisting of audio with the same frequency as the first expected value data but a different period, and to determine whether the meter buzzer sound of the object being measured is triggered by the first expected value data or the second expected value data through the following process: In the pre-processing stage, The meter buzzer being measured receives a signal of either the first or second expected value data, and the resulting meter buzzer sound is recorded by a microphone and captured as audio data. The recording data is processed to match the length of the first expected value data to become the first processed recording data, and the recording data is processed to match the length of the second expected value data to become the second processed recording data. The first processed audio data and the second processed audio data are each signal-processed to reduce noise. In the determination unit, The first correlation coefficient between the first expected value data and the first processed recording data is obtained. A second correlation coefficient was obtained between the second expected value data and the second processed recording data. Based on the threshold values ​​of the first and second correlation coefficients, a determination is made as to whether the first or second processed recording data matches the first expected value data, matches the second expected value data, or is different from the first and second expected values.

[0008] The automatic detection method for meter buzzer sounds that solves the above problem is: When inputting a signal to sound a confirmation buzzer to the meter buzzer being measured, and determining whether the meter buzzer sound emitted from the meter buzzer matches the designed expected audio data, The signal for either the first or second expected value data is input to the meter buzzer being measured, and the sound of the meter buzzer that sounds is acquired as recording data by a microphone. The recorded data is processed to match the length of the first expected value data to obtain the first processed recorded data, and the recorded data is processed to match the length of the second expected value data to obtain the second processed recorded data. The first processed recording data and the second processed recording data are each subjected to signal processing to reduce noise. Obtain the first correlation coefficient between the first expected value data and the first processed recording data. We obtain a second correlation coefficient between the second expected value data and the second processed recording data. Based on the threshold values ​​of the first and second correlation coefficients, a determination is made as to whether the first or second processed recording data matches the first expected value data, matches the second expected value data, or is different from the first and second expected values. [Effects of the Invention]

[0009] According to the present invention, when comparing audio data of meter buzzers with the same frequency but different periods, it is possible to provide an automatic meter buzzer sound determination device and an automatic determination method that can determine more easily and reliably than in the conventional method whether or not the meter buzzer sound conforms to the design value. [Brief explanation of the drawing]

[0010] [Figure 1] It is a diagram showing the configuration of the first embodiment of the meter buzzer sound automatic determination device according to the present invention. [Figure 2] It is a flowchart for explaining the automatic determination method of the meter buzzer sound according to the present invention. [Figure 3] It shows data obtained by recording the meter buzzer sound at the first level and performing determination, where (A) is the first expected value data at the first level, (B) is the processed recording data obtained by processing the recording data at the first level to the length of the first expected value data at the first level, (C) is the processed recording data obtained by processing the recording data at the first level to the length of the second expected value data at the second level, and (D) is the second expected value data at the second level. [Figure 4] It shows data obtained by recording the meter buzzer sound at the second level and performing determination, where (A) is the first expected value data at the first level, (B) is the processed recording data obtained by processing the recording data at the second level to the length of the first expected value data at the first level, (C) is the processed recording data obtained by processing the recording data at the second level to the length of the second expected value data at the second level, and (D) is the second expected value data at the second level.

Mode for Carrying Out the Invention

[0011] Hereinafter, the present invention will be described in detail based on several embodiments shown in the drawings. However, each embodiment is only for explaining the present invention, and the present invention is not limited thereto. [First Embodiment] FIG. 1 is a diagram showing the configuration of the first embodiment of the meter buzzer sound automatic determination device 10 according to the present invention. As shown in FIG. 1, the meter buzzer sound automatic determination device 10 is, for example, a device that automatically determines whether the initial expected value of the meter buzzer sound of a newly designed vehicle matches the sound that actually occurs when, for example, the incorporated meter buzzer is driven with the data of the expected value. That is, the meter buzzer sound automatic determination device 10 inputs the designed expected value audio data to the meter buzzer to be measured, and determines whether the meter buzzer sound emitted from the meter buzzer matches the expected value audio data. The meter buzzer sound automatic detection device 10 comprises a control unit 11, a storage unit 12, a microphone 13, an input unit 14, a display unit 15, a pre-processing unit 16, and a determination unit 18.

[0012] The control unit 11 is a so-called CPU and can be composed of a microprocessor, microcontroller, etc. Multiple software programs necessary for the operation of the meter buzzer sound automatic detection device 10 are stored in the memory unit 12. The CPU and memory unit 12 control each part of the meter buzzer sound automatic detection device 10. The memory unit 12 is composed of DRAM, non-volatile memory such as a hard disk drive (HDD) or flash memory (SSD), etc.

[0013] Furthermore, the program that enables the automatic meter buzzer sound detection device 10 is recorded and stored on a storage medium readable by a computer consisting of a CPU and a memory unit 12, etc. A CD-ROM, DVD-ROM, USB memory, etc., can be used as the storage medium. Additionally, the program may be stored in the memory unit 12 via a network connected to the automatic meter buzzer sound detection device 10.

[0014] Microphone 13 is used to record the sound produced by a meter buzzer installed in an actual vehicle. Microphone 13 may be connected to an amplifier or an AD converter that converts analog signals to digital signals, and may also be connected to a CPU via an I / O interface. The audio data of the meter buzzer acquired by microphone 13 is called the recorded data.

[0015] Furthermore, for playback of recorded data, the meter buzzer sound automatic detection device 10 may be configured to include a speaker. The recorded data played back from the speaker, as well as the initially expected designed audio data described later, are processed as digital signals within the CPU. These are then connected to a DA converter that converts the digital signals to analog signals, and further connected to the speaker via an amplifier and an I / O interface.

[0016] The input unit 14 consists of a keyboard and mouse used by the operator of the meter buzzer sound automatic detection device 10.

[0017] The display unit 15 shows a screen for the operator to monitor program execution, etc. The display unit 15 may be a display that also functions as a keyboard and mouse, such as a touch panel. In this case, the input unit 14, which consists of a keyboard and mouse, may be omitted.

[0018] The preprocessing unit 16 processes the expected value data (described later) and the waveform of the recorded audio data of the meter buzzer acquired by the microphone 13, performing waveform processing to reduce noise. Waveform processing for noise reduction includes bandwidth limiting using a bandpass filter and waveform processing such as trimming.

[0019] The determination unit 18 calculates the correlation between the audio data designed as the initial expected value and the meter buzzer recording data acquired by the microphone 13, and determines whether the designed audio data, which is the expected value, matches the meter buzzer recording data acquired by the microphone 13. Specifically, it makes a determination by comparing whether the correlation coefficient between the expected audio data and the recording data exceeds a predetermined value, i.e., a threshold.

[0020] (Method for automatically detecting meter buzzer sound) Next, as an example, a method for automatically determining a meter buzzer sound using the automatic meter buzzer sound determination device 10 according to the present invention will be described. The automatic meter buzzer sound determination method of the present invention is characterized by its ability to more easily determine data with the same frequency but different periods, which was previously difficult to distinguish. As an example of data with the same frequency but different periods, two buzzer sounds related to not fastening a seat belt will be used to explain. In other words, the device is configured to automatically determine whether the meter buzzer sound to be measured is a first expected value data consisting of a sound with a predetermined frequency and a predetermined period, as designed expected value sound data, or a second expected value data consisting of a sound with the same frequency as the first expected value data but a different period.

[0021] (Two buzzer sounds indicating that the seat belt is not fastened) The first meter buzzer sound (hereinafter referred to as the first level meter buzzer sound or simply Lv1) is the initial warning sound when the seat belt is not fastened. The second meter buzzer sound (hereinafter referred to as the second level meter buzzer sound or simply Lv2) is the warning sound that occurs after a certain period of time has elapsed since the initial warning sound for when the seat belt is not fastened. The first level meter buzzer sound and the second level meter buzzer sound have the same tone but different periods. The second level meter buzzer sound has a shorter period than the first level meter buzzer sound.

[0022] Figure 2 is a flowchart illustrating the automatic meter buzzer sound determination method according to the present invention. In the flowchart of Figure 2, the determination of three meter buzzer sounds is explained as an example, consisting of the first level meter buzzer sound, the second level meter buzzer sound, and another meter buzzer sound with a different frequency from the first level and second level meter buzzer sounds.

[0023] As shown in Figure 2, when the meter buzzer sound determination starts, in step ST1, the name of the expected buzzer sound to be sounded is assigned as an arbitrary variable (step ST1). In step ST2, it is determined whether the meter buzzer sound to be determined is a seat belt (Yes) or not (No), and if the target is a seat belt (Yes), the process proceeds to step ST3. On the other hand, in step ST2, if the item to be judged is not a seat belt (No), the process ends.

[0024] (Judgment of seatbelt sound (Lv1 and Lv2)) In step ST3, a signal for sounding the meter buzzer to be measured is input, and the sounded meter buzzer sound is recorded by the microphone, that is, the sounded meter buzzer sound is acquired as recording data by the microphone. Further, this recording data is processed so as to match the length of the first expected value data to obtain first processed recording data. In the following description, it is assumed that a signal for sounding the meter buzzer to be measured is input, and the sounded sound is different in frequency from the first expected value data, the second expected value data, the meter buzzer sound of the first level, and the meter buzzer sound of the second level (also referred to as other meter buzzer sounds). Therefore, in step ST3, the recording data acquired by the microphone is any one of the sound according to the first expected value data, the sound according to the second expected value data, and other meter buzzer sounds.

[0025] Specifically, the recording data is shaped (trimmed) so as to match the length of the first expected value data (master) of the first meter buzzer sound (Lv1). This recording data is called first processed recording data that is processed so as to match the length of the first expected value data (also referred to as the expected value data of Lv1). Specifically, the recorded data (for example, 10-second data) is trimmed to the length of the first expected value data (for example, 3-second data). At this time, which part of the recording data is trimmed can be determined by performing cross-correlation and aligning the peak positions so as to determine the start position. Thereby, the first processed recording data becomes 3-second data.

[0026] Both the first expected value data and the recording data are functions of the time axis, and each data is represented by X = {x 1, x 2, ‥‥, x n} and Y = {y 1, y 2, ‥‥, y n}. In this case, the correlation coefficient between X and Y is the pair of two data X = {x 1, x 2, ‥‥, x n} and Y = {y 1, y 2, ‥‥, y nAssuming that}, and with Vx and Vy being the variances of X and Y respectively, Sx and Sy being the standard deviations, and Sxy being the covariance, the correlation coefficient r is expressed by the following equation (1).

[0027]

number

[0028] In step ST4, the recording data acquired in step ST3 is processed to match the length of the second expected value data (master) for the second meter buzzer sound (Lv1). This recording data is called the second processed recording data, which has been processed to match the length of the second expected value data. Specifically, the recording data is trimmed to match the length of the second expected value data (master) for the second meter buzzer sound (Lv2). This recording data is called the second processed recording data, which has been processed to match the length of the second expected value data (also called the expected value data for Lv2). Specifically, the recorded data (e.g., 10 seconds of data) is trimmed to the length of a second expected value data (e.g., 6 seconds of data). The starting position for trimming can be determined by cross-correlation, aligning the peak positions. As a result, the processed second recorded data becomes 6 seconds long.

[0029] In step ST5, the audio data consisting of the first expected value data and the trimmed first processed recording data acquired in step ST3, and the audio data consisting of the second expected value data and the trimmed second processed recording data acquired in step ST4, are subjected to preprocessing for noise reduction. As part of the noise reduction preprocessing, low and high frequencies are removed and normalized using a bandpass filter on the audio data. For example, the frequency range of the bandpass filter can be from 200Hz to 3000Hz.

[0030] In step ST6, (A) the correlation coefficient between the first expected value data obtained and preprocessed in step ST5 and the trimmed first processed recording data is obtained. When obtaining the correlation coefficient, five moving averages may be calculated and the average taken.

[0031] In step ST7, (B) the correlation coefficient between the second expected value data obtained and preprocessed in step ST5 and the processed second processed recording data is obtained. When obtaining the correlation coefficient, five moving averages may be calculated and the average taken.

[0032] In step ST8, it is determined whether the correlation coefficient of (B) obtained in step ST7 is 0.7 or higher (Yes) or not (No). If the correlation coefficient of (B) is not 0.7 or higher (No), the process proceeds to step ST9, where it is determined that the recorded data is different from the seatbelt buzzer sound. Then, in step ST13, the data obtained in steps ST6 and ST7, namely the first expected value data, the second expected value data, and the trimmed first and second processed recorded data, are stored in the output folder as evidence, and the determination is completed (step ST14).

[0033] On the other hand, if the correlation coefficient for (B) is 0.7 or higher (Yes) in step ST8, proceed to step ST10.

[0034] In step ST10, it is determined whether the correlation coefficient of (A) obtained in step ST6 is 0.7 or higher (Yes) or not (No). If the correlation coefficient of (A) is not 0.7 or higher (No), the process proceeds to step ST12, where it is determined that the second processed recording data is the second seatbelt buzzer sound. Then, in step ST13, the data obtained in steps ST6 and ST67, namely the first expected value data, the second expected value data, and the trimmed first and second processed recording data, are stored in the output folder as evidence, and the determination is completed (step ST14).

[0035] On the other hand, in step ST10, if the correlation coefficient of (A) is 0.7 or higher (Yes), the process proceeds to step ST11, where it is determined that the first processed recording data is the Level 1 buzzer sound of a seat belt. Then, in step ST13, the data acquired in steps ST6 and ST7, namely the Level 1 expected value data, the Level 2 expected value data, and the trimmed first and second processed recording data are stored in the output folder as evidence, and the determination is completed (step ST14).

[0036] According to the meter buzzer sound automatic determination device 10 of the present invention, when comparing audio data with the same frequency but different periods, the determination of the meter buzzer sound can be performed more easily than in the conventional method.

[0037] According to the automatic meter buzzer sound determination method of the first embodiment of the present invention, a signal is input to the meter buzzer to be measured to sound a buzzer sound to be confirmed, and when determining whether the meter buzzer sound sounded by the meter buzzer matches the designed expected value sound data, in order to distinguish between the first expected value data and the second expected value data which has a similar frequency to the first expected value data, the meter buzzer sound sounded by the meter buzzer is recorded to obtain recording data, this recording data is processed into time-axis data of the length of the first expected value data to obtain the first processed recording data, the recording data is processed into time-axis data of the length of the second expected value data to obtain the second processed recording data, a first correlation coefficient is obtained between the first processed recording data and the first expected value data, and a second correlation coefficient is obtained between the second expected value data and the second recording data, and the meter buzzer sound is determined based on the threshold of these correlation coefficients, thereby improving the determination accuracy compared to conventional methods.

[0038] Therefore, when comparing the conventionally difficult-to-distinguish meter buzzer sounds produced by the first expected value data and the second expected value data, the meter buzzer sound can be determined more easily and with higher accuracy than before. In other words, when comparing the expected value of audio data with the same frequency but different periods with recorded data, a comparison in the frequency domain is not possible, so a comparison in the time domain is performed. Specifically, the recorded data is processed to the same length as the expected value data, the processed recorded data is compared with the expected value data, and the determination is made using the thresholds of the first and second correlation coefficients described above, thereby improving the determination accuracy. This makes it easier to determine sounds that were previously difficult to distinguish, such as the first and second expected value data. The following provides a more detailed explanation with reference to an embodiment. [Examples]

[0039] The configuration of the automatic meter buzzer sound detection device 10 is shown below. Computer name: Mouse Computer (Model number: B4-i7) CPU: Intel (Model: Core i7-1260P) OS: Windows 10 Storage: DRAM: 32GB SSD: 512GB Software used for calculations: Microsoft Excel (registered trademark) *Program language (size): Pythone (56kB)

[0040] The first meter buzzer sound, the second meter buzzer sound, and the meter buzzer sound with a different frequency were identified according to the flowchart in Figure 2. Figure 3 shows data obtained by recording and judging the meter buzzer sound at the first level. (A) is the first expected value data at the first level, (B) is the processed recording data obtained by processing the recording data at the first level to the length of the expected value data at the first level, (C) is the processed recording data obtained by processing the recording data at the first level to the length of the second expected value data at the second level, and (D) is the second expected value data at the second level.

[0041] In Figure 3, the correlation coefficient between the first expected value data of the first level shown in (A) and the first processed recording data of the first level shown in (B) was 0.7 or higher (see steps ST6 and ST10 in Figure 2). Furthermore, the correlation coefficient between the trimmed second-level processed recording data shown in Figure 3(C) and the second expected value data shown in Figure 3(D) was 0.7 or higher (see steps ST7 and ST8 in Figure 2). Based on this, the recording data was determined to be the first expected value data of the first level.

[0042] Figure 4 shows data obtained by recording and judging the second meter buzzer sound, where (A) is the first expected value data of the first level, (B) is processed recording data obtained by processing the second level recording data to the length of the first expected value data of the first level, (C) is processed recording data obtained by processing the second level recording data to the length of the second expected value data of the second level, and (D) is the second expected value data of the second level.

[0043] In Figure 4, the correlation coefficient between the first expected value data of the first level shown in (A) and the second recording data of the first level shown in (B) was 0.5, which is less than or equal to 0.7 (see ST7 and ST9 in Figure 2).

[0044] Furthermore, the correlation coefficient between the trimmed second-level second recording data shown in Figure 4(C) and the second expected value data shown in Figure 4(D) was 0.7 or higher (see steps ST7 and ST8 in Figure 2). Based on this, the recording data was determined to be the second-level second expected value data.

[0045] The present invention can be implemented in various forms without departing from its spirit. The automatic meter buzzer sound determination device 10 of the present invention can make determinations even in a moving vehicle, not just a stationary vehicle, as long as recorded data can be acquired. In this case, the expected value data can be stored in the storage unit 12 as an expected value database.

[0046] For example, in the embodiment described above, the case in which the first expected value data is processed was explained, but of course, the processing of the expected value data is adjusted appropriately in response to the meter buzzer sound in order to improve the judgment accuracy. [Explanation of Symbols]

[0047] 10. Automatic detection device for meter buzzer sound 11 Control Unit 12 Storage section 13 Mike 14 Input section 15 Display section 16 Pre-processing section 18 Judgment section

Claims

1. An automatic meter buzzer sound determination device that inputs a signal to sound a confirmation buzzer sound to a meter buzzer under measurement, and determines whether the meter buzzer sound sounded by the meter buzzer matches the designed expected value sound data, The automatic determination device comprises a control unit, a storage unit, a microphone, an input unit, a pre-processing unit, and a determination unit. The control unit, The system is configured to create the aforementioned expected value audio data as first expected value data consisting of audio at a predetermined frequency and period, and second expected value data consisting of audio with the same frequency as the first expected value data but a different period, and to determine whether the meter buzzer sound of the object being measured is triggered by the first expected value data or the second expected value data through the following process: In the aforementioned pre-processing unit, The signal of the first expected value data or the second expected value data is input to the meter buzzer of the object to be measured, and the sound of the meter buzzer that sounds is acquired as recording data by the microphone. The aforementioned recording data is processed to match the length of the first expected value data to obtain the first processed recording data, and the aforementioned recording data is processed to match the length of the second expected value data to obtain the second processed recording data. The first processed recording data and the second processed recording data are each subjected to signal processing for noise reduction. In the determination unit, A first correlation coefficient is obtained between the first expected value data and the first processed recording data. A second correlation coefficient is obtained between the second expected value data and the second processed recording data. An automatic meter buzzer sound determination device that determines, based on the threshold of the first correlation coefficient and the threshold of the second correlation coefficient, whether the first or second processed recording data matches the first expected value data, matches the second expected value data, or is data different from the first expected value and the second expected value.

2. An automatic meter buzzer sound determination method using the automatic meter buzzer sound determination device described in claim 1, In an automatic meter buzzer sound determination method, a signal is input to the meter buzzer to be measured to sound a confirmation buzzer sound, and the method determines whether the meter buzzer sound sounded by the meter buzzer matches the designed expected value sound data, The signal of the first expected value data or the second expected value data is input to the meter buzzer of the object to be measured, and the sound of the meter buzzer that sounds is acquired as recording data by a microphone. The aforementioned recording data is processed to match the length of the first expected value data to obtain the first processed recording data, and the aforementioned recording data is processed to match the length of the second expected value data to obtain the second processed recording data. The first processed recording data and the second processed recording data are each subjected to signal processing to reduce noise. Obtain a first correlation coefficient between the first expected value data and the first processed recording data. The second correlation coefficient between the second expected value data and the second processed recording data is obtained. An automatic meter buzzer sound determination method, which determines whether the first or second processed recording data matches the first expected value data, matches the second expected value data, or is data different from the first expected value and the second expected value, based on the threshold of the first correlation coefficient and the threshold of the second correlation coefficient.

Images